CoolFace
Modelpublic

shopifyinterngrinder/sidekick-autocomplete-06b-sft-real

sourceHugging Faceapache-2.0updated 6mo agoView on Hugging Face
0likes11downloads
Model Card

shopifyinterngrinder/sidekick-autocomplete-06b-sft-real

Fine-tuned from Qwen/Qwen3-0.6B using TRL SFT.

Training Details

ParameterValue
Base ModelQwen/Qwen3-0.6B
Datasetshopifyinterngrinder/sidekick-autocomplete-data-real @ main
Training Examples13,565
Validation Examples1,508
Epochs3
Learning Rate2e-05
Batch Size (per device)1
Gradient Accumulation2
Max Sequence Length512
Precisionbf16
Optimizeradamwtorchfused
Warmup Steps50
Weight Decay0.01
LR Schedulercosine
PackingEnabled
Dataset Formatchat

Framework Versions

LibraryVersion
Transformers4.57.6
TRL0.29.0
PyTorch2.8.0+cu128
Datasets3.6.0
Accelerate1.13.0